Advanced monitoring and control of anaerobic wastewater treatment plants: fault detection and isolation

Abstract

In this paper, a fault detection and isolation approach using fuzzy logic is described for on-line analysis of problems occurring in anaerobic digestion processes. The measurements available on the process are preprocessed to build a vector of fault residuals indicating the magnitude of the problems. This vector is classified into a prespecified category (i.e., a class) which is a state of the system, according to discrimination fuzzy rules. Three different types of classes were defined in a hierarchical structure : sensors faults, sub-process faults and process faults. This approach was developed to handle in real time both technical and biological problems. Demonstration of the practical interest of this study was made using real life experiments and large improvement of the reliability and safety of the process was obtained, thus optimizing the overall wastewater treatment.